Culture medium, FBS, and Tryp-LE were from Invitrogen. Collagenase type II and trypsin were from Worthington Biochemical Corporation. Trametinib was from Cayman Chemicals. All other chemicals were from Fisher unless otherwise noted. All animal procedures were approved by Cornell University’s Institutional Animal Care and Use Committee (IACUC), protocols 2005 − 0151 (clinical sample collection), 2017-0035 (experimental mouse work), and 2023-0034 (clinical trial).
Tumor sample collection
After owners consented to sample collection, dogs were placed under general anesthesia. Two tumor samples were collected, with a 4 mm punch device for diagnostic use, and with a 2 mm punch device for deposit with the Cornell Veterinary Biobank (CVB). When the tumor was later surgically excised as part of the dog’s treatment, it was delivered to the Cornell University Progressive Development of Therapeutics (PATh) facility to generate PDX models. All procedures conducted on dogs were in accordance with accepted best practices and AVMA guidelines, and in accordance with approved Cornell University’s IACUC (protocol 2005 − 0151), and no extraordinary steps were conducted to obtain samples for use in these experiments.
Histology
Tumor samples were embedded in formalin, and 4 µm slices were taken. These slices were mounted to charged slides, and then samples were processed with an automated IHC stainer largely as previously described 62. Paraffin was removed with Bond dewax solution (Leica), and then the samples were exposed to Bond epitope retrieval solution (Leica) for 30–40 minutes. Samples were then incubated with anti AE1/AE3 (DakoCytomation, #M3515, to detect cytokeratin 2) or anti-MIB-1 (DakoCytomation, #M7240, to detect Ki67), followed by alkaline phosphatase conjugated anti-mouse IgG (Leica, #PV6110) and RedDetection CM (Leica, #DS9390). Samples were alternately stained with hematoxylin and eosin stain. Slides were scanned with an Aperio CS2 ScanScope (Vista). All histological analyses were conducted by board-certified pathologists blinded to the condition being examined.
Patient-derived xenograft development
NOD.Cg-PrkdcscidIl2rgtm1Wjl/SzJ (NSG) mice were bred in-house by the Cornell PATh facility, and were thus fully acclimated before experiments began. Tumor samples were collected from dogs as described above. Samples ~ 2–3 cm3 were placed in DMEM with penicillin and streptomycin for transport. Within 2–3 hours, the samples were minced into 2–3 mm3 pieces. Several samples were cryopreserved, and the remaining samples were divided evenly and implanted into five NSG mice (female, 6–8 weeks old). Mice were anesthetized with 2.5% isofluorane, and pieces were implanted into the left flank. Mice were administered analgesics as needed for pain and monitored carefully for two hours following surgery. Once tumors reached ~ 100 mm3, approximately 36 weeks after implantation, the mouse was sacrificed using carbon dioxide euthanasia (3.5 L/min), the tumor was harvested, minced into segments, and the segments were implanted into new mice (male or female as available, no more than 10 weeks old). Samples were also cryopreserved, and other samples were used to develop cell culture models as described below.
Cell line development
Cell lines were isolated from tumors by adapting a procedure from a study on breast cancers 47. Briefly, tumor samples were collected from PDX-carrying mice, or directly from dogs. Tumors were soaked in 70% ethanol for ~ 30 seconds, and then minced with sterile blades. The tumors were washed twice with PBS to remove blood, and then transferred to a sterile conical tube. 5 mL of digestion buffer (2.5 mg/mL trypsin, 5 mg/mL albumin, 850 units/mL collagenase type II in PBS) was added to the tumor samples. The tube was capped and placed in a 37°C shaker and shaken at 550 RPM for 20 minutes. The mixture was then filtered through a cell strainer, and the tumor pieces captured by the strainer were transferred to a clean tube, suspended in 5 mL of fresh digestion buffer, and shaken for another 20 minutes. To this solution, 10 mL of wash medium (F12 medium supplemented with 5% FBS and 250 ng/mL gentamicin) was added. The tube was spun down (250 X g for 10 minutes), and 10 mL of medium (expected to contain primarily fibroblasts) was removed. The remaining 5 mL of medium, and tumor pieces, were expected to contain primarily epithelial cells.
Epithelial cells were collected via gravimetric enrichment. To the tube containing 5 mL of medium and tumor pieces, 10 mL of wash medium was added. The tube was inverted several times to suspend cells, and the cells were allowed to settle for 20 minutes. 10 mL of medium was removed and spun down to collect a cell pellet, which was suspended in RPMI-1640 supplemented with 10% FBS and dispensed into a single well of a 6-well plate. This ‘wash + collect + spin’ procedure was repeated 12 more times on the tube with tumor pieces, to create 13 culture samples. The tumor pieces were then suspended in RPMI 1640 + 10% FBS and dispensed into a well plate as well. Cells were kept highly confluent for 1–2 weeks, after which we began passaging them with greater dilution. In general, we found that each subsequent wash resulted in more homogeneous looking cells, but that cells in the high-wash wells were sometimes not confluent enough to grow successfully. We made use of the highest wash-number sample which showed robust growth to perform further experiments.
Clonogenic colony formation assay:
Cells were grown to ~ 70% confluence, and then passaged using Tryp LE. Cells in suspension were counted, and 1000 cells were dispensed into a 10 cm dish. The cells were grown for 4 weeks, or until significant colonies were readily visible on inspection, whichever occurred first. The cells were then washed three times with PBS, fixed to the plates with 3.7% formaldehyde in PBS, and stained with 4% crystal violet in methanol for 5 minutes. The crystal violet was removed, and the cells were washed three times with water (5 minutes each wash) then one final water wash (2 hours). The plated colonies were then photographed.
Cell drug dose studies
Cells at ~ 70% confluence were washed with PBS and exposed to Tryp-LE solution for ~ 5 minutes. The cells were then suspended in RPMI-1640 supplemented with 10% FBS and transferred to a sterile conical tube. The cells were counted, then dispensed into 96-well plates at a density of 1000 cells per well. The cells were allowed to settle overnight, and the following day the medium was removed and replaced with medium containing the indicated DMSO-solvated drugs at assorted concentrations, or with DMSO alone as a control. The medium was again replaced on the fourth day of culture. On the sixth day of culture, the medium was removed, replaced with fresh medium, and cell viability was determined with Presto Blue reagent (Thermo). Viability was determined colorimetrically, following ~ 1–2 hours incubation with the reagent, on a Tecan Spark microplate reader. IC50 values were determined in GraphPad Prism using a two-parameter, variable slope logistic curve.
Western blotting: Cells were grown to ~ 70% confluence, washed 3 times with PBS, and then lysed in lysis buffer (20 mM HEPES pH 7.6, 150 mM NaCl, 1 mM EDTA, and 1% NP-40). Protein concentration in the lysate was determined with BioRad Protein Assay Dye per the manufacturer’s instructions. 30 µg of protein in Laemmli buffer was then loaded onto 4–20% Tris-glycine gels and resolved via SDS-PAGE. Protein was transferred to PVDF membrane, and the membrane was blocked overnight in milk. The membrane was rinsed with TBST and stained for one hour with CypA antibody (1:1000 dilution in TBST, Cell Signaling #51418), phosphor-ERK antibody (1:2000 dilution, Cell Signaling #9106), or vinculin antibody (1:2000 dilution, Cell Signaling #13901). The membrane was washed 4X with TBST, and then exposed for two hours to HRP-linked anti-rabbit IgG (Cell Signaling #7074, 1:5000 dilution in TBST) or anti-mouse IgG (Cell Signaling #7076, 1:5000 dilution). The membrane was washed 4X with TBST, exposed to Western Lighting Plus Chemiluminescence reagent (Perkin Elmer), and imaged on a Bio-Rad ChemiDoc. Band density was quantitated using ImageJ 63.
Genotyping analysis: DNA was collected from dog tumor, PDX tumor, and cell line samples using the Zymo DNA micro-prep kit per the manufacturer’s instructions. DNA was then submitted to Illumina for genotyping on the EMBARK version of the CanineHD 220k bead array. Quality control on the genotype data was performed in PLINK 1.9 (www.cog-genomics.org/plink/1.9/) 56. Samples with missingness values of > 3% were considered to have failed and were removed. For the remaining samples, SNPs with any missingness were removed, and then linked SNPs were removed, resulting in 13,178 informative SNPs being maintained. For each sample pair, the IBD was calculated using a metric called pi-hat (^π), which is a measure of the probability that the two samples share 0, 1 or 2 alleles. To generate a ‘fingerprint’ for each cell line, we selected 168 SNPs across the genome to differentiate the cell-lines from each other. Criteria for SNP selection included: no missingness, located > 10Mb apart, minor allele frequency (MAF) > 30% in the samples for the present study, and MAF > 30% in a group of 880 dogs genotyped on the same 220k array for other studies. Further manual selection was done to maximize genotype differences between the cell-lines. To show the relationship between each of the samples for these 168 SNPs, a heatmap was generated using the package pheatmap v 1.0.12 64 in R 65. To determine the uniqueness of the SNPs chosen for fingerprinting, pairwise IBD was calculated using only the 168 SNPs for all samples in this study and the 880 dogs that have been previously genotyped.
In vivo drug study
Trametinib was dissolved in vehicle (10% Cremophor EL, 10% PEG400, 80% PBS). Co-F-1236 cells (2 X 106) were suspended in Matrigel, and then injected into the left flank of male NSG mice (8–10 weeks old). Drug treatments began once the majority of mice had palpable tumors. Mice were randomly sorted into drug-treatment or vehicle-treatment groups. Mice were weighed daily to determine proper treatment volume. Six mice were treated with the indicated amount of trametinib daily by oral gavage, and six were treated with an equivalent volume of vehicle alone. Tumors were measured 3 times per week with calipers. Tumor volume was estimated as 0.5 * length * width * width. Representative mice were subjected to MRI before drug treatment began, and immediately prior to sacrifice. MRI images were analyzed using VivoQuant Imaging Software v3.5 by Invicro. At the experimental endpoint, mice were euthanized with carbon dioxide (3.5 L/min), then tumors were harvested, photographed, and immediately weighed before significant desiccation could occur. Data for tumor growth were analyzed for statistical significance using a two-tailed Student’s t-test.
Ongoing clinical study
Client-owned dogs with a confirmed diagnosis of COSCC arising from an oral mucosal surface are enrolled in an 8-week study to evaluate the response and tolerability of trametinib. Eligibility requirements include completely staged patients that have no evidence of regional or distant metastasis, have not received previous treatment of the oral tumor (i.e., surgical excision, chemotherapy or radiation therapy), are considered suitable candidates to undergo general anesthesia, are not diagnosed with other debilitating or chronic systemic diseases, are not pregnant or lactating, and the owners can comply with the required follow up appointments during the study. All enrolled patients receive an oral dose of trametinib (0.015–0.02 mg/kg once daily) during the study. Response to therapy is monitored via a weekly phone call to document potential side effects observed at home by the owner, biweekly physical and oral examination including caliper measurements of visible oral mass and routine bloodwork (i.e., complete blood cell count and serum biochemistry panel), and full oral tumor staging under general anesthesia at days ~ 28 and ~ 56, including a contrast-enhanced (2ml/kg Iohexol 350mg iodine/ml) computed tomography (CT) head exam (16-slice Aquilion LB; Canon Medical Systems USA, Inc, Tustin, CA), diagnostic imaging of the abdominal cavity (CT or ultrasound) and thorax (CT or 3-view radiographs), and cytological or histological assessment of regional lymph nodes. The head CT exams are acquired in continuous helical axial slices, with slice thickness ranging from 0.5mm to 2.0 mm and reconstructed into volumes (effective slice thickness 0.3mm to 1.0 mm). The CT studies are reviewed by a board-certified veterinary radiologist as DICOM studies in the hospital’s Picture Archiving and Communication system [PACS] (Carestream VuePACS; Rochester, NY, USA) on medical diagnostic-quality monitors (Dell U3219Q; Dell Technologies, Round Rock, TX, USA). Tumor volume (cm3) is calculated as the sum of the hand-drawn cross-sectional area (cm2) of the abnormal [tumor] tissue on each axial slice, multiplied by the effective slice thickness. Tumor volume is determined on each head CT exam (timepoints: enrollment, d. 28, and d.56) and compared to document changes between each timepoint. All clinical procedures are done in accordance with a protocol (2023-0034) reviewed and approved Cornell University’s IACUC.